A Tale of Two Classes: Adapting Supervised Contrastive Learning to Binary Imbalanced Datasets

2025-03-21

AI Summary

Supervised contrastive learning (SupCon) suffers significant performance degradation in binary-class long-tailed scenarios—such as medical diagnosis—due to representation collapse. To address this, we propose an interpretable local neighborhood distribution assessment framework, introducing two novel local representation quality metrics that uncover structural deficiencies invisible to conventional global metrics. Guided by these insights, we design two new contrastive strategies: weighted positive/negative sample sampling and adaptive temperature scaling—constituting the first SupCon optimization paradigm explicitly tailored for binary-class imbalanced settings. Extensive experiments across seven natural and medical image datasets demonstrate that our method improves downstream classification accuracy by up to 35% over standard SupCon.

Technology Category

• Machine Learning: Semi-Supervised Learning
• Computer Vision: Representation Learning for Vision
• Search and Optimization: Local Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Supervised contrastive learning (SupCon) has proven to be a powerful alternative to the standard cross-entropy loss for classification of multi-class balanced datasets. However, it struggles to learn well-conditioned representations of datasets with long-tailed class distributions. This problem is potentially exacerbated for binary imbalanced distributions, which are commonly encountered during many real-world problems such as medical diagnosis. In experiments on seven binary datasets of natural and medical images, we show that the performance of SupCon decreases with increasing class imbalance. To substantiate these findings, we introduce two novel metrics that evaluate the quality of the learned representation space. By measuring the class distribution in local neighborhoods, we are able to uncover structural deficiencies of the representation space that classical metrics cannot detect. Informed by these insights, we propose two new supervised contrastive learning strategies tailored to binary imbalanced datasets that improve the structure of the representation space and increase downstream classification accuracy over standard SupCon by up to 35%. We make our code available.

Problem

• Adapting supervised contrastive learning for binary imbalanced datasets
• Addressing performance decline in SupCon with class imbalance
• Improving representation space structure for better classification accuracy

Innovation

• Adapts supervised contrastive learning for binary imbalance
• Introduces novel metrics for representation space quality
• Proposes strategies improving accuracy by up to 35%